djyde

We all have a day when we'll be over 70, I wonder if you'll still think the same way then?

This isn't a personality issue at all—it's pure disrespect. If someone treated me like that, I wouldn't befriend them or open up to them either. Sincerity is a two-way street.

If your open source code contains your technical barriers, then don't open source the code with those barriers—only open source the other parts.

If it's the other case, where you're worried about plagiarism, I actually don't think you need to be too concerned. I once saw an interview with Airbnb's founder Brian Chesky where he talked about how Airbnb also faced many imitators in its early days. These competitors grew rapidly too, but looking back, the difference between imitators and originals is that while imitators might get off to a quick start, they find it hard to persevere through difficulties like the original does in the mid-to-late stages. In the end, it's often the original who stuck with their vision that survives.

Same

On Earth Day, I just launched my own Hacker News app that I really love! https://haiker.app!

My favorite is "Do things that don't scale" (https://www.paulgraham.com/ds.html), which has been really inspiring for my indie product work. Especially for those of us building internet products, we often tend to build scalable things right from the start, but in reality, products in the early stages require founders to do a lot of non-automated dirty work.

For entrepreneurs, reading Paul Graham's articles is incredibly useful.

(Sent via haiker.app — My handmade Hacker News app)

The scariest part is that humans are starting to use AI to generate spam comments, which in turn get used to train the models. Will the language capabilities of these models just keep getting worse?

Sent via haiker.app — My handmade Hacker News app

Hi Hackers,

I’m a heavy Hacker News user and a non-native English speaker. To be honest, I often find it struggling to browse through threads and even harder to participate because English isn't my first language.

So I wrote an app called Haiker. It can automatically translate HN stories and comments. When I want to reply, I can type in my native language and translate it into English with one tap. This has helped me become much more active in discussions and discover a lot of valuable content.

I believe many other non-native hackers have the same need.

That said, it’s not exclusively for non-native speakers. It’s a beautiful HN client in its own right. Feel free to give it a try!

What level of programming tasks can a 27B model handle? Even with Claude, I'm occasionally not satisfied, and I can't imagine how effective a 27B model would be.

me too btw :P

The first thing many people do after installing OpenClaw is summarize Hacker News

I suspect they've implemented some kind of intelligent email filtering system

题外话，你们不觉得在 openclaw 里用 claude 相当浪费 token 吗？

I'm really curious how he manages to read through so many emails every day.

While I don't agree with many things Cook has done during his tenure, like the Touch Bar and removing the SD card slot from MacBooks, I have to admit the man knows how to make money.

My approach is to try delegating as many daily tasks to AI as possible. For example, I've now completely replaced my read-it-later tool with my agent. I just send him links, and he periodically summarizes the content I've sent him.

Too many tokens, and high-level languages have better semantics, making them more suitable for reasoning.

Generally, it's automatically included when Claude Code makes a commit.

But I've recently found that Cursor's composer2 is also really good to use.

I always thought Craig would become CEO.

I've been using glm5.1 for pretty much all my coding work, but Claude is too expensive for me. Haven't tried qwen yet though. China's coding models are now very cost-effective.

Many language learning extension have this feature. They provide a transcript view beside the player.

Actually, the nice thing about chrome extensions is that users can view all requests made by content_script and background script through devtool. This is much more intuitive than checking the code.

I understand your concern. I think "access to much information" means this extension require <all_url> host_permission, which I don't want to either but it must.

Because custom AI provider's API base url is submit by user. If I want to call the API on background script, this base url must be listed on host_permissions. Otherwise it will cause a CORS problem.

optional_host_permissions may fix this problem, but since the base url is set by user, it's not possible to use this workaround.

Any suggestion?

Sorry, no. I'm an indie developer so I wish I can make some money with it. It would be a fair price as I wrote on landing page, will less than $20/year.

But I can share the tech stack I use:

- WXT: the browser extension framework

- Vercel AI SDK. Both SDK core and SDK UI. I wrote a bridge that I can proxy the api request to background script which compat with the SDK streaming protocol, so that I can use the SDK UI.

- Next.js: for the landing page and the up coming cloud service (e.g. prompt sync)

- Strapi: headless cms.

Hi hackers. I wrote an AI browser extension for chat with page.

I know there are many existing tools like this. But every existed tools require me login before using it and I must use their AI models (which is the way they make money).

So I write my own tools for it. The key features are:

- No need to login, just use it

- Use your own AI API key

- Use your own custom prompt

- Chat run entirely within the browser, without passing through any third-party servers

I wrote a chrome extension for this a year ago https://dengtab.com/

Wonderful. I'm interested in learning about circuit and PCB design. What would you recommend as the best introductory resources to get started?

No offense, but I'm curious if this landing page is AI generated? I don't know why it feels a bit AI-generated.

Thanks for the heads up! I thought that was the full amount of data.

My idea is to write a purely front-end application that takes a json file and uses a chart to visualize some interesting information by year, such as the most watched channels for the selected year